COVID-19 Neural Net

A critical step in the fight against COVID-19 is effective detection of infected patients so that those infected can receive immediate treatment and care, as well as being isolated to mitigate the spread of the virus. The main detection method used to detect COVID-19 cases is the reverse transcriptase-polymerase chain reaction (RT-PCR).

An alternative detection method that has also been used for the detection of COVID-19 has been the radiography examination, where radiologists take and analyze chest radiography images (for example, CXR chest radiograph or CT scan) to look for visual indicators associated with COVID-19.

We used a pre-trained MobileNetV2 and data augmentation, as the original dataset COVID-19 CT segmentation dataset is very small. We also used bagging to see if we could improve the performance of the model.

The code with more explanations is in my GitHub Pattern-Recognition-COVID.